PRSM

A desktop and mobile client for OpenClaw. Built with Electron, React, TypeScript, and Capacitor.

OpenClaw is an open-source AI assistant platform that lets you run, customize, and orchestrate AI agents on your own infrastructure.

PRSM connects to your OpenClaw gateway over WebSocket and gives you a native interface to chat with agents, manage skills and cron jobs, edit workspace files, and record voice notes — on Windows, Linux, and Android.

Chat interface with session sidebar and skills panel

Grab the latest release from GitHub Releases.

• Real-time streaming — responses render token-by-token as the agent thinks
• Image attachments — paste from clipboard, drag-and-drop, or pick from file system (auto-compressed to stay under the 25 MB WebSocket limit)
• Voice notes — record audio, transcribe via any Whisper-compatible STT service, insert as text
• Pinned messages — star important messages; pins persist locally across sessions
• Full markdown rendering — fenced code blocks with syntax highlighting, tables, links, lists
• Thinking mode toggle — enable extended thinking; reasoning steps render as compact, collapsible blocks
• Ctrl+Enter to send — on desktop, Enter inserts a newline; Ctrl+Enter sends the message

• Multi-session management — create, switch, rename, and delete chat sessions
• Name sessions on creation — optionally label a new chat when you start it
• Session search/filter — search sessions by name in the sidebar

• Agent profiles — view and edit agent workspace files (SOUL.md, MEMORY.md, ACTIVE-WORK.md, etc.)
• Multi-agent switching — switch between agents from the sidebar

• Skills browser — browse all installed skills with status badges (Ready / Missing / Disabled)
• Skill details — view documentation, required binaries, enable/disable
• Cron job viewer — inspect scheduled tasks, job configuration, and run history

• Dark & light themes — toggle manually or follow system preference
• Desktop notifications — get notified when an agent responds in the background
• Secure token storage — credentials encrypted via Electron's safeStorage API
• Heartbeat filtering — background heartbeat messages hidden from the chat feed
• Auto-updater — check for updates from the sidebar; configurable update policy
• Mobile-optimized UI — responsive layout with slide-out panels, mobile keyboard handling, and touch-friendly controls
• Cross-platform — Windows, Linux, and Android from a single codebase

Download the artifact for your platform from the Releases page.

• Windows Installer — run the .exe, follow the wizard.
• Windows Portable — just run the .exe from any folder.
• Linux — chmod +x PRSM-*.AppImage && ./PRSM-*.AppImage
• Android — enable "Install from unknown sources," then open the APK.

1. Open PRSM — on first launch it shows the connection dialog.
2. Enter your OpenClaw Gateway URL (e.g., ws://192.168.1.50:18789 or wss://your-server.com).
3. Enter your Gateway Token or Password (depending on your gateway's auth.mode).
4. Click Save & Connect.

Start a new session, pick an agent, and go. Messages stream in real-time. Use the right panel to browse skills, view cron jobs, or manage pinned messages.

Voice notes let you record audio and have it transcribed to text using any Whisper-compatible speech-to-text (STT) service. Click the microphone button in the input area to configure — or click "Ask your assistant for help" and the agent will walk you through it.

Three fields to fill in:

docker run -d --gpus all -p 8000:8000 \
  ghcr.io/speaches-ai/speaches:latest-cuda

docker run -d -p 8000:8000 \
  ghcr.io/speaches-ai/speaches:latest

docker run -d -p 8080:8080 localai/localai:latest

# Clone
git clone https://github.com/CRMbyRSM/PRSM.git
cd PRSM

# Install dependencies
npm install

# Development (Vite hot reload + Electron)
npm run dev

Every release MUST include ALL four packages. Use build:release — never build or build:linux alone.

# ✅ CORRECT — builds everything: Windows NSIS + Portable, Linux AppImage, Android APK
npm run build:release

# ❌ WRONG — only builds one platform, not a complete release
npm run build
npm run build:linux

The build:release script:

1. Builds Electron app for Windows (NSIS installer + portable) and Linux (AppImage)
2. Builds the mobile web bundle
3. Syncs to the Capacitor Android project
4. Compiles the Android APK via Gradle
5. Copies all artifacts to release/

First-time Android setup (if android/ directory doesn't exist):

export ANDROID_HOME=/home/riktanius/android-sdk
export ANDROID_SDK_ROOT=/home/riktanius/android-sdk
npx cap add android    # only needed once — generates the android/ project
npm run build:release

See RELEASE.md for the full release checklist.

npm run build:win        # Windows (Setup + Portable)
npm run build:linux      # Linux (AppImage)
npm run build:mac        # macOS (DMG)
npm run mobile:sync      # Build web assets + sync to Android
npm run mobile:android   # Open in Android Studio

PRSM/
├── electron/              # Electron main process
│   ├── main.ts                # App lifecycle, auto-updater, safeStorage
│   └── preload.ts             # Context bridge (IPC to renderer)
├── src/
│   ├── App.tsx                # Root component, routing
│   ├── main.tsx               # Entry point
│   ├── components/
│   │   ├── ChatArea.tsx           # Message list, markdown rendering, thinking blocks
│   │   ├── InputArea.tsx          # Message input, attachments, voice recording
│   │   ├── Sidebar.tsx            # Session list, search, agent selector, update button
│   │   ├── TopBar.tsx             # Connection status, theme toggle
│   │   ├── RightPanel.tsx         # Skills, cron jobs, pinned messages tabs
│   │   ├── SettingsModal.tsx      # Connection + STT configuration
│   │   ├── VoiceSettings.tsx      # STT provider setup
│   │   ├── AgentDetailView.tsx    # Agent profile and workspace file editor
│   │   ├── SkillDetailView.tsx    # Skill docs and status
│   │   ├── CronJobDetailView.tsx  # Cron job config and run history
│   │   ├── SubagentBlock.tsx      # Sub-agent activity display
│   │   ├── UpdateBanner.tsx       # Update notification banner
│   │   ├── CertErrorModal.tsx     # TLS certificate error handling
│   │   └── ErrorBoundary.tsx      # React error boundary
│   ├── store/
│   │   └── index.ts           # Zustand state (sessions, settings, pins — persisted to localStorage)
│   ├── lib/
│   │   ├── openclaw-client.ts # WebSocket client for the OpenClaw gateway protocol
│   │   ├── platform.ts        # Platform abstraction (Electron / Capacitor / web)
│   │   └── safe-render.ts     # XSS sanitization for rendered content
│   ├── styles/
│   │   └── index.css          # CSS with custom properties for dark/light theming
│   └── test/
│       └── setup.ts           # Vitest test setup
├── android/                   # Capacitor Android project
├── scripts/                   # Build and test utilities
├── screenshots/               # README screenshots
└── prsm-icons/                # App icons (all platforms)

The app communicates with the OpenClaw gateway over a single persistent WebSocket connection. All messages, sessions, skills, cron data, and agent workspace files flow through this connection using OpenClaw's native protocol.

npm test              # Watch mode
npm run test:run      # Single run
npm run test:coverage # With coverage report
npm run typecheck     # TypeScript type checking
npm run lint          # ESLint

PRSM is forked from ClawControl by Jacob L. Edwards / Oaken Cloud Technologies, LLC. The original project provided the foundation that PRSM builds on.

Built for the OpenClaw ecosystem. Join the community on Discord.

Developed and maintained by RSM Consulting — crmbyrsm.com.

MIT — see LICENSE for details.